Which customer segment should a startup win first?

  1. Blog
  2. Strategy & Positioning

By Saroj Jha · November 2, 2026 · 15 min read

Win first the segment where five things are true at once: the pain is urgent, you can reach the buyers, they can pay, they talk to each other, and you can deliver the whole product they need today. Score every candidate segment on those five, drop any that scores a 1 on one of them, check that your revenue target needs a believable share of the winner, and then stay there until referrals and a shorter sales cycle tell you the segment is working for you. Only then move to the neighbouring segment that can reuse what you built.

In three numbers

  • 43% of 385 VC-backed startups that shut down since 2023, with a known cause, cited poor product-market fit[1]
  • 22% to 33% Superhuman’s “very disappointed” score after it segmented to focus on users like its very disappointed group[2]
  • 58% the same score after three quarters of work to improve the product[2]

The case for a beachhead is a case about focus, so the rest of this article is about choosing where to focus and when to stop.

Running case: Example Co. Fictional

Example Co. is a fictional seed-stage startup with six people and a mobile app that turns on-site safety inspections into finished compliance reports. It has 14 paying customers, spread across fire-safety inspectors, HVAC contractors, elevator maintenance firms and a couple of property managers, because the founders said yes to anyone who asked. Its revenue target is $1.5 million in annual recurring revenue within 24 months, and its two founders do all the selling. Every number about Example Co. in this article is invented for illustration.

What is a beachhead segment, and why start with one?

A beachhead is the first segment you set out to dominate, chosen because winning it gives you the references, revenue and know-how to take the segments next to it. The idea comes from Geoffrey Moore’s Crossing the Chasm (Harper Business, first published in 1991, third edition 2014), which argues that a start-up selling something new should become the clear leader in one chosen niche, building out the whole product for it, before going after the mainstream market[3]. The image is a military landing: take one stretch of beach with overwhelming force, then move inland.

Bill Aulet turned the idea into a working step in MIT’s entrepreneurship teaching. The MIT course material for 15.390 New Enterprises lists the questions to ask of each candidate segment before choosing a beachhead[4], and Aulet describes the beachhead as the market where, once you hold a dominant share, you have the strength to attack adjacent markets[5]. His short version is to pick the one segment where you have the highest odds of success and that gives you the legitimacy and assets to grow further[6].

The reason to start narrow is that pragmatic buyers want proof from companies like their own. A startup with 14 customers in four industries has, in effect, three or four references in each, and none of them persuades the next buyer as well as 30 references in one industry would. The 43% figure in the key band is not proof that narrow focus prevents failure, because CB Insights coded causes from public post-mortems and interviews of companies that had already shut down[1]. It is a reminder that “no market that wants this enough” is the most common product reason startups die, and a beachhead is the cheapest way to find out whether one segment wants it.

Example Co., step 1 Fictional

Of the 14 customers, 6 are fire-safety inspection firms, 4 are HVAC contractors, 2 are elevator maintenance firms and 2 are property managers. The founders’ first instinct is to keep all four, because each brought revenue. My view is that four segments with two to six customers each is four half-built products, so the first job is to choose one.

Do this

Write down every segment your current customers belong to, by industry and by the job the buyer holds, before you decide anything.

What makes a segment worth winning first?

I use five criteria, each a question you can answer with evidence. They are close to the MIT list[4], which also asks about entrenched competition and whether results can be shown within one to two years, and they share Moore’s insistence on the whole product[3].

Urgent pain. The buyer has a problem they are already trying to solve, with money or with hours, and something bad happens if they do not. MIT’s version is whether the customer has “a compelling reason to buy”[4]. A problem the buyer agrees with but has never tried to fix is not urgent.

Reachable. You can get in front of these buyers with the team you have. That means you can name the associations, events, communities, publications or marketplaces where they gather, and a founder can book ten conversations in a fortnight. MIT asks whether the target customer is “readily accessible to our sales force”[4].

Can pay. The buyer has a budget line, or can create one, at a price that works for you. MIT asks whether the customer is “well funded”[4]. A segment that loves the product but cannot pay more than your cost to serve them is a community, not a market.

Word of mouth within the segment. Buyers in the segment talk to each other, through trade bodies, regional groups, shared suppliers or plain gossip. This is the criterion founders skip, and it is the one that makes a beachhead compound. Aulet’s description of a beachhead market includes that word-of-mouth communication exists between its customers, along with similar buying and similar sales cycles[5].

A whole product you can deliver. You can give these buyers everything they need to get the result, including integrations, onboarding, compliance documents and support hours, either yourself or with partners. MIT phrases it as whether you can “today, with the help of partners, deliver a whole product to fulfill that reason to buy”[4]. A segment that needs a feature you plan to build next year is a future segment.

Example Co., step 2 Fictional

The founders list what they know about each segment. Fire-safety inspectors face a hard legal deadline for every report and gather at a state association’s annual conference. Property managers like the app but have no habit of talking to each other about inspection tools, and they need integrations with three property platforms Example Co. does not have.

Do this

For each candidate segment, write one line of evidence per criterion; if you cannot write the line, the score is a guess.

Figure 1 Five questions decide the first segment, and each has a source
  • Criterion The question Source wording
  • Urgent pain Is the buyer already spending money or hours on the problem, with something bad happening if they don’t? “a compelling reason to buy”[4]
  • Reachable Can a founder book ten conversations in a fortnight, through places these buyers gather? “readily accessible to our sales force”[4]
  • Can pay Is there a budget line, or one that can be created, at a price that works for you? “well funded”[4]
  • Word of mouth Do buyers in the segment talk to each other? “Word of mouth communication exists between customers in the market”[5]
  • Whole product Can you deliver everything they need to get the result today, yourself or with partners? “deliver a whole product” today, with the help of partners[4]; Moore’s “building out the whole product”[3]

MIT OpenCourseWare 15.390, Class Six slides[4]; Bill Aulet, MIT Sloan Executive Education[5]; Geoffrey Moore’s book page for Crossing the Chasm[3]. The questions in the middle column are my wording of the criteria in the article.

How do you score candidate segments without fooling yourself?

Score each candidate from 1 to 5 on each criterion, average the five, and convert the average to a 0 to 100 scale so a segment that scores 1 everywhere gets 0 and one that scores 5 everywhere gets 100. Equal weights are the honest default. If you change the weights, decide them before you score, not after you see which segment you secretly prefer.

Two rules keep the scoring honest. First, a 1 on any criterion rules the segment out, whatever its average, because a segment you cannot reach or that cannot pay will not be rescued by strong pain. Second, every score needs a line of evidence from a conversation, a public source or your own data. Where the evidence is missing, lower the score by one point and see whether the winner changes; if it does, the next task is to collect that evidence, not to start selling.

My rule of thumb, not a benchmark: the winner should lead the runner-up by at least 10 points on the 0 to 100 scale. A narrower lead means the choice is close enough that one more round of customer conversations is cheaper than a wrong year. The customer interviews guide covers how to run those conversations without leading the buyer.

Example Co., step 3 Fictional

Scores are pain, reach, can pay, word of mouth, whole product. Fire-safety inspection firms: 5, 4, 4, 5, 4; total 22, average 4.4, score (4.4 - 1) / 4 x 100 = 85. Elevator maintenance firms: 5, 2, 5, 4, 3; total 19, average 3.8, score 70. HVAC contractors: 4, 3, 3, 3, 4; total 17, average 3.4, score 60. Property managers: 3, 5, 4, 1, 2; total 15, average 3.0, score 50, but ruled out by the 1 on word of mouth. Fire-safety leads elevator maintenance by 85 - 70 = 15 points.

Do this

Score your segments in a shared sheet with the evidence column filled in, then ask a cofounder to lower every score that lacks evidence and see if the winner holds.

Figure 2 Fire-safety inspection firms lead by 15 points, and property managers drop out on one score
  • Fire-safety inspection firmsScores 5, 4, 4, 5, 4 Leads by 15 points
  • Elevator maintenance firmsScores 5, 2, 5, 4, 3
  • HVAC contractorsScores 4, 3, 3, 3, 4
  • Property managersScores 3, 5, 4, 1, 2 Ruled out: word of mouth scored 1
Show the numbers as a table
Segment Pain Reach Can pay Word of mouth Whole product Average Score Status
Fire-safety inspection firms 4.4 In
Elevator maintenance firms 3.8 In
HVAC contractors 3.4 In
Property managers 3.0 Ruled out

Example Co. (fictional). Each segment scored 1 to 5 on pain, reach, can pay, word of mouth and whole product; score = (average − 1) ÷ 4 × 100, so (4.4 − 1) ÷ 4 × 100 = 85, (3.8 − 1) ÷ 4 × 100 = 70, (3.4 − 1) ÷ 4 × 100 = 60, (3.0 − 1) ÷ 4 × 100 = 50; 85 − 70 = 15. A 1 on any criterion rules a segment out. Bars to scale from zero on the 0 to 100 scale.

Is the segment big enough, or too big?

A segment can score well and still be the wrong first choice, because it is too small to carry your next round or too large to dominate. Aulet’s check is to estimate the revenue per year you would earn with 100% of the beachhead, so you can see whether it is neither too big nor too small[6]. I add a second check that founders find more useful: divide your revenue target by your annual contract value to get the number of customers you need, then divide that by the number of reachable accounts in the segment.

My rule of thumb, not a benchmark: if the target needs more than about a quarter of the reachable accounts, the segment is too small to be your only one, because nobody wins most of a market in two years. If it needs under 2 or 3%, the segment may be too broad to dominate, and you should look for the sub-segment that talks to itself. Market sizing for investors is a different exercise, covered in How to size your market; this check is only about whether the first segment can carry the first target.

Example Co., step 4 Fictional

Target $1.5 million ARR. Fire-safety: annual contract value $9,000, so 1,500,000 / 9,000 = 166.7, rounded up to 167 customers, out of 1,200 reachable firms: 167 / 1,200 = 13.9%. Elevator maintenance: $15,000 contract value, 100 customers needed out of 300 firms: 33.3%, too small to carry the target alone. Fire-safety passes the size check; elevator maintenance becomes a later segment, not the first.

Do this

Count reachable accounts from a list you could actually contact (an association directory, a licensing register, a marketplace), not from a top-down market report.

Figure 3 The leading segment needs 13.9% of its accounts; the runner-up would need a third

Fire-safety inspection firms Score 85, the leader

  • Contract value $9,000 a year
  • Customers needed 167
  • Reachable firms 1,200
  • Share needed 13.9%
  • Result Passes the size check

Elevator maintenance firms Score 70, the runner-up

  • Contract value $15,000 a year
  • Customers needed 100
  • Reachable firms 300
  • Share needed 33.3%
  • Result Too small to carry the target alone

Share of reachable accounts needed, to scale out of 100%

Fire-safety inspection firms167 of 1,200 13.9% Elevator maintenance firms100 of 300 33.3% A third Show the numbers as a table
Segment Contract value Customers needed Reachable firms Share needed
Fire-safety inspection firms $9,000 167 1,200 13.9%
Elevator maintenance firms $15,000 100 300 33.3%

Example Co. (fictional), against a $1.5 million ARR target. $1,500,000 ÷ $9,000 = 166.7, rounded up to 167; 167 ÷ 1,200 = 13.9%. $1,500,000 ÷ $15,000 = 100; 100 ÷ 300 = 33.3%. My rule of thumb, not a benchmark: above about a quarter, a segment is too small to be the only one.

What does winning the beachhead look like in practice?

Once the segment is chosen, everything visible should say it. Your homepage, case studies, pricing examples, conference budget and integrations roadmap point at one buyer, so that a prospect in that segment sees their own world and a prospect outside it understands they are not the focus yet. How to turn that choice into a positioning statement is the subject of Startup positioning: how to become the obvious choice, and I will not repeat it here.

Saying no is the hard part. Keep serving customers you already have in other segments, but stop marketing to those segments and stop building features only they need. In my view, the moment to hold the line is when a large prospect from outside the segment asks for a custom feature; taking it trades the compounding of one segment for a single contract.

The Superhuman numbers in the key band show what narrowing can reveal. Rahul Vohra, the company’s founder, reported that 22% of the 100 to 200 users polled said they would be very disappointed without the product, that the share was 33% after the company segmented to focus on users like its very disappointed group, and that it reached 58% after three quarters of work to improve the product[2]. These are one company’s self-reported figures on a single survey question, not a controlled study, and as I read Vohra’s account, the 33% came from segmenting the users Superhuman already had rather than from new customers. What they illustrate is that an average across all users can hide a segment where the product already works.

Example Co., step 5 Fictional

Example Co. rewrites its homepage for fire-safety inspection firms, books a stand at the state association’s conference, and moves the property-platform integrations off the roadmap. The 8 customers outside the segment keep their service, but no new marketing goes to their industries.

Do this

List the three things you will stop doing for segments outside the beachhead, and tell your team in writing.

When should you move to the next segment?

Move when the beachhead starts selling itself, not when you get bored of it. I watch three signals together, and my thresholds below are rules of thumb, not benchmarks. A rising share of new deals arrives by referral or from buyers who already heard of you inside the segment; the sales cycle shortens because references do the persuading; and your win rate in the segment holds or rises as you sell to buyers who are less like your early fans.

When two of the three signals hold for two consecutive quarters, start the next segment. In my view each segment you win should make the next one easier to take. The next segment is usually the segment that shares the most with the first: the same whole product with small changes, or the same word-of-mouth network, or the same buyer role in a neighbouring industry. Check the same product-market fit signals in the new segment before spending on it, because fit does not transfer automatically; Product-market fit: how to tell if you have it sets out those signals.

The opposite mistake is staying too long. If the beachhead is won and growth is flat because you have reached most of the reachable accounts, holding the line becomes a ceiling. The size check from earlier tells you roughly when that will happen.

Example Co., step 6 Fictional

Fifteen months in, Example Co. has 52 fire-safety customers, 52 / 1,200 = 4.3% of reachable firms. Of the last 40 new deals, 18 came by referral or from conference contacts: 18 / 40 = 45%. The median sales cycle has fallen from 74 days to 41 days, (41 - 74) / 74 = a 44.6% reduction. Two signals hold for two quarters, so the founders open HVAC contractors next, because HVAC firms buy similar inspection reports and many share fire-safety firms’ customers.

Do this

Put the three signals on one line of your monthly report, with the beachhead and all other segments shown separately.

Figure 4 Two of three expansion signals held by month 15
  1. Month 0 Fire-safety inspection firms chosen 6 of 14 paying customers are in the segment; 8 elsewhere keep their service
  2. Month 15 52 fire-safety customers 52 of 1,200 reachable firms: 4.3%
  3. Signal 1 Held: 45% of new deals by referral or conference contacts 18 of the last 40 new deals
  4. Signal 2 Held: median sales cycle 44.6% shorter From 74 days to 41 days
  5. Signal 3 Win rate in the segment Not one of the two signals that held
  6. Next Open HVAC contractors Similar inspection reports, and many share fire-safety firms’ customers
Show the numbers as a table
Measure Value
Customers in the segment, month 0 6 of 14
Customers in the segment, month 15 52 (4.3% of 1,200)
New deals by referral or conference contacts 18 of last 40 (45%)
Median sales cycle 74 to 41 days (44.6% shorter)

Example Co. (fictional), step 6. 52 ÷ 1,200 = 4.3%; 18 ÷ 40 = 45%; (41 − 74) ÷ 74 = −44.6%. Two of the three signals held for two quarters.

Figure 5 One company’s score rose after it segmented its users, and again after three quarters of product work
  • All users polled100 to 200 users 22%
  • After segmentingto focus on users like its very disappointed group 33%
  • After three quarters of work to improve the product 58%
Show the numbers as a table
Point Very disappointed
All users polled (100 to 200) 22%
After segmenting to focus on users like its very disappointed group 33%
After three quarters of work to improve the product 58%

Rahul Vohra, Superhuman’s founder, in First Round Review, 2018[2]: share of users who said they would be “very disappointed” without the product. One company’s self-reported figures on a single survey question, not a controlled study; only the first poll has a stated sample. Bars to scale out of 100%.

What changes by stage?

  • Pre-seed — You probably have no segment data yet. Shortlist three segments, run ten conversations in each, and score them only on evidence from those conversations; expect to change your mind.
  • Seed — This is the usual moment to choose, when you have 10 to 30 customers spread too thin. Run the scoring and the size check, and accept the short-term revenue cost of saying no.
  • Series A–B — You should be on your second or third segment. The question becomes sequencing: which adjacent segment reuses the most of your whole product and references, and whether each new segment gets its own messaging and pipeline numbers.

Free tool

Score your segments before you pick one

The Beachhead Segment Scorer scores two to ten candidate segments on pain, reach, budget, competition and proof, applies the rule-out on any 1, runs the size check against your revenue target, and tells you whether the pick still wins when unevidenced scores drop by a point.

Open the scorer

Choosing the segment is the strategic decision; turning it into a list of accounts and a segment plan is the operational one. The Customer Segmentation Playbook covers how to define and size segments in depth, and the ICP and Account-Scoring Playbook turns the winning segment into a scored list your team can work from.

If you want me to look at your own funnel, positioning and channels, my Growth Audit ends with three prioritized fixes and a 30-day action plan.

Related on AAJ

Articles

Playbooks

Templates and tools

References

The superscript numbers in the article match this list.

Every source was opened on November 1, 2026, through a web-page reading tool, so check short quotations against the live documents.

  1. CB Insights (2026). The top 9 reasons startups fail. 5 March 2026. Vendor study: 431 VC-backed companies that shut down since 2023, excluding those with a prior exit, coded from public post-mortems, founder interviews and shutdown announcements; reasons identified for 385 companies; companies could have several reasons. cbinsights.com
  2. Vohra, R. (2018). How Superhuman built an engine to find product/market fit. First Round Review, 12 November 2018. Founder’s account; “very disappointed” share from a survey of 100 to 200 users; segmented and later figures self-reported without sample sizes. review.firstround.com
  3. Moore, G. A. (1991; 3rd edition 2014). Crossing the Chasm. Harper Business. Book page on the author’s site. geoffreyamoore.com
  4. MIT OpenCourseWare (2013). 15.390 New Enterprises, Spring 2013, Class Six: Target customer definition (slides credited to Bill Aulet, Howard Anderson and Matt Marx). ocw.mit.edu
  5. MIT Sloan Executive Education (2014). Launching a successful start-up 3: the beachhead market (Bill Aulet). Post dated 21 January 2014. executive.mit.edu
  6. Aulet, B. (2024). Disciplined entrepreneurship: 6 questions for startup success. MIT Sloan Ideas Made to Matter, 16 April 2024. mitsloan.mit.edu

More in Strategy & Positioning

Part of the Strategy & Positioning hub - see all 24 resources on this topic.

Also useful in Audience & Customer Research.